atlassian-remote

Official
GitHub Repo

Overview

The atlassian-remote MCP Server is a cloud-hosted Model Context Protocol (MCP) implementation from Atlassian that lets AI assistants and agents securely access and interact with Atlassian Cloud data — including Jira, Confluence, and Compass — without manual integrations.

Transport

sse

Select Tools

  • atlassianUserInfo
  • getAccessibleAtlassianResources
  • getConfluenceSpaces
  • getConfluencePage
  • getPagesInConfluenceSpace
  • getConfluencePageFooterComments
  • getConfluencePageInlineComments
  • getConfluencePageDescendants
  • createConfluencePage
  • updateConfluencePage
  • createConfluenceFooterComment
  • createConfluenceInlineComment
  • searchConfluenceUsingCql
  • getJiraIssue
  • editJiraIssue
  • createJiraIssue
  • getTransitionsForJiraIssue
  • transitionJiraIssue
  • lookupJiraAccountId
  • searchJiraIssuesUsingJql
  • addCommentToJiraIssue
  • getJiraIssueRemoteIssueLinks
  • getVisibleJiraProjects
  • getJiraProjectIssueTypesMetadata

Key Capabilities

  • Enterprise data access – Query and retrieve Atlassian context from Jira, Confluence, and Compass in real time, without switching apps.
  • Natural language actions – Create and update issues, pages, and components from conversational commands via AI assistants.
  • Summarization and insight – Summarize long Jira work items or extensive Confluence documentation through AI probes with contextual awareness.
  • Cross-tool workflows – Execute multi-step operations (e.g., bulk issue creation or linking documentation to tasks) in a single conversational flow.
  • Secure and permission-aware – Uses OAuth 2.1 authentication and respects existing Atlassian permissions so tools only see and act on data the user is allowed to access.

How It Works

Instead of running locally, the atlassian-remote MCP Server is hosted by Atlassian at a cloud endpoint (e.g., https://mcp.atlassian.com/v1/mcp) and exposes Atlassian product operations over HTTP/SSE-based MCP transport. Clients connect by configuring their MCP settings to reference this remote server URL, and then initiate a secure browser-based OAuth 2.1 authorization flow that grants scoped access to Jira, Confluence, and other Atlassian Cloud products.

Once authenticated, supported AI tools stream contextual requests and receive structured responses in real time. When an agent invokes a tool (like searching issues or creating a Confluence page), the remote MCP server calls the corresponding Atlassian Cloud APIs on behalf of the user, returning results or confirmation over the MCP protocol.

For desktop or local IDE clients (e.g., VS Code, Cursor, Claude Desktop), a lightweight mcp-remote proxy can be used to bridge the local MCP client to the hosted server endpoint, triggering the OAuth login and relaying MCP calls through the remote server.